Machine learning (ML) has advanced predictive modeling in medical diagnosis and risk assessment through large clinical datasets, yet applications for predicting post-cancer depression and anxiety remain limited. This cross-institutional, longitudinal study aims to develop ML models to predict depression and anxiety in cancer patients and to identify key contributing factors.

ML models were developed and validated to predict depression and anxiety disorders within one year of six cancer diagnoses across three medical centers in Taiwan, from January 2017 to December 2022 (n = 24,580). Study variables included demographic, clinical, and quality of care attributes. ML algorithms employed include Logistic Regression (LR), Random Forest (RF), K-Nearest Neighbor (KNN), Adaptive Boosting (AdaBoost), and Extreme Gradient Boosting (XGBoost). Model performance was evaluated using accuracy, precision, recall, F1 score, and area under the receiver operating characteristic (AUROC) curve. Statistical analysis was conducted with SPSS 26.0 and Spyder 3.8 (Python).

The XGBoost model significantly outperformed others, achieving 98.30% accuracy, 98.17% precision, 98.30% F1 score, and 99.72% AUROC (P < 0.001). Feature importance analysis identified tumor size, age, and body mass index as the top risk factors for depression and anxiety disorders within one year of a cancer diagnosis.

ML algorithms advance the understanding of depression and anxiety in cancer patients by leveraging longitudinal data to identify key predictive factors. These models not only enhance mental health care and treatment quality but also provide insights that inform evidence-based guidelines, thereby improving outcomes for patients, families, and healthcare providers during cancer care.

Our hospital is an urban academic multisite facility in Southwestern Ontario. The General Adult Ambulatory Mental Health Service (GAAMHS) delivers acute urgent and non-urgent outpatient (O-P) psychiatric care for adults 18 to 64 years. In the context of sub-optimal physician resources and the COVID-19 pandemic, there was an accumulation of 812 non-urgent referrals in March 2021. Manual review of the number of incoming referrals and processing timelines estimated a wait time of 9 to 12 months to see a psychiatrist. This quality improvement project was conducted to resolve the backlog of referrals and to reduce the wait times for the incoming non-urgent referrals.

This project was developed and implemented by the core team of a programme manager, an administrative assistant (AA) and a psychiatrist. It was achieved without any additional funding for project management. Process mapping of various components of GAAMHS was completed and an Ishikawa diagram was created to identify factors contributing to the backlog. Quality improvement change ideas were proposed and tested using Plan-Do-Study-Act cycles. The interventions included reassessment of patient needs, implementation of an electronic data capture tool and team-based model of care, refining the referral triage process and standardising the service delivery practices of psychiatrists.

The 812 backlogged referrals were resolved in a median of 5.3 months. The average number of new consultations and total O-P visits per full-time psychiatrist per month was 10.3 ±3.6 and 74.5±15.9, respectively, in 2020; it increased to 17.1±7.9 and 80.8±21.6 in 2022 and 18.8±8.9 and 90.0±20.2 in 2023. The wait times for the new incoming referrals have continued to decrease with the median wait times in December 2023 being 102 days and the wait times for the 75th percentile being 145 days.

A combination of strategies helped resolve the backlog and reduce wait times to access acute O-P mental health.

School-based health centers (SBHCs) and other school-based clinics (OSBCs) reduce health care access barriers and support positive outcomes in disadvantaged children by providing primary medical care and other services, respectively. This study describes California SBHCs and OSBCs and identifies school characteristics associated with access.

Databases on California SBHCs, OSBCs, and schools were compiled. Descriptive statistics characterize the number of clinics and services offered from 2011 to 2023. Multivariable logistic regression models estimate associations between school characteristics and SBHC/OSBC access in 2023.

Between 2011and 2023, the SBHCs increased from 124 to 186, and services became more comprehensive. OSBCs increased from 18 to 104, with most offering mental health services. SBHC access was more likely in schools with a larger share of students who were English Learners and "other" as their race and ethnicity; high schools; and in large cities. OSBC access was more likely in schools with a larger share of English Learner students and in cities.

SBHCs and OSBCs expanded and largely served disadvantaged schools, which may promote health and academic equity.

The Connecting Adolescents to Comprehensive Healthcare (CATCH) program built upon existing infrastructure for school nurses and physicians to provide limited reproductive health services to New York City public high school students. We evaluated CATCH reach, service delivery, and impacts on contraceptive use and pregnancy among female teens over the period 2011-2019.

Our evaluation incorporated data from school rosters and CATCH patient records; the NYC Youth Risk Behavior Survey, to estimate contraceptive use among students without CATCH access for comparison with students with access; and NYC vital statistics, to estimate the pregnancies, abortions, and births averted by CATCH.

CATCH grew from piloting on five campuses to operating on 61 campuses with more than 80,000 students, reaching an estimated 53.7% of sexually active female students on those campuses by the 2018-19 school year. Use of most or moderately effective contraception (IUD, implant, pills, patch, ring, or Depo-Provera) among CATCH patients increased over time and was consistently higher than estimates for those same students if they had not had CATCH access. We estimate the program averted 3526 pregnancies among NYC teens.

By supporting access to reproductive health care, CATCH contributed to contraceptive uptake and reduced pregnancies among NYC teens.

Close to half of those engaged with community mental health teams (CMHT) report an alcohol or drug problem. UK public health guidance recommends that these services screen for harmful alcohol use, but reporting may be less likely amongst minoritised ethnic groups. This study aimed to explore: (i) the prevalence of screening and referrals to alcohol services within CMHTs and differences across ethnic groups; (ii) how alcohol use is assessed and treated in CMHTs, and tailored for minoritised ethnic service users; and (iii) staff and minoritised ethnic service users' experiences of assessing and reporting alcohol use.

A rapid appraisal was conducted which triangulated data across patient healthcare records (aim 1), online survey (aim 2), interviews and focus groups (aim 3) with three CMHT services within an NHS Mental Health Foundation Trust in North-West England. Data was analysed using framework analysis.

Both patient notes and survey data showed that alcohol was seldom assessed using formal tools. Three themes were developed reflecting differences in the barriers of reporting and assessing alcohol use for minoritised ethnic service users and staff. With barriers for the former including information sharing and barriers for the latter including protecting the therapeutic relationship.

Triangulating data from across different sources highlights the complex challenges that services face in meeting the recommendations around alcohol screening in CMH services. Our findings have implications on the need for staff in mental health services to better understand and accommodate the needs of minoritised ethnic service users who may have co-occurring alcohol and mental health problems.

Studies have identified differences by patient characteristics in addiction treatment utilization in the early COVID-19 pandemic period, yet an understanding of longitudinal changes in utilization patterns remains unclear. We examined treatment utilization trends over three years post-pandemic, with a particular focus on differences by age and race and ethnicity.

Using electronic health record data, this retrospective cohort study examined overall and telehealth addiction treatment initiation and engagement 3 years pre- and 3 years post-pandemic (3/16/2020) following identification of 170,618 episodes involving problematic substance use among 124,413 adults in a large, integrated Northern California health system. Interrupted time series models were fit to examine annual utilization rates during pre- and post-pandemic periods (3/1/2017-1/15/2020 and 5/17/2020-2/28/2023, respectively), and level- and trend-changes in utilization from pre- to post-pandemic, overall and by age group and race and ethnicity.

Overall treatment initiation decreased from 27.0% to 24.2% during the pre-pandemic period by approximately 2% annually (RR [95% CI] = 0.98 [0.96, 0.99]), increased by 6% after the onset of the pandemic (1.06 [1.03, 1.10]), and then decreased by 2% annually in the post-pandemic period to 23.0% (0.98 [0.79, 0.99]). Telehealth initiation increased from 1.9% to 2.6% during the pre-pandemic period by 12% annually (RR [95% CI] = 1.12 [1.06, 1.19]), increased five-fold immediately after pandemic onset (RR [95% CI] = 5.14 [4.62, 5.72]), and then decreased by 10% annually (RR [95% CI] = 0.90 [0.88, 0.92]). Overall and telehealth engagement followed similar patterns. Pre- to post-pandemic trends in utilization varied by age group and slightly by race and ethnicity, which may have been primarily driven by initial increases in utilization at the onset of the pandemic.

Following immediate increases in treatment initiation and engagement during the pandemic, utilization via telehealth decreased slightly over time. Availability of telehealth was not associated with increased or sustained utilization over time. Despite some variation in trends over time by age group and race and ethnicity, we did not find strong evidence of differences across groups.

Evidence on neighborhood food stores and public health is limited. This study examined the association of availability of neighborhood food stores and health and well-being.

We used three-wave data (2013, 2016, and 2019) from a nationwide cohort study of physically and cognitively independent older adults ≥65 years old in Japan. Exposure was the perceived availability of neighborhood food stores in 2016. We assessed 40 health/well-being outcomes in 2019 across seven domains; physical/cognitive health, health behaviors, mental health, psychological well-being, social well-being, character and virtue, and cognitive social capital. We adjusted for pre-exposure covariates including prior outcome in 2013. We included 47,318 respondents for 4 outcomes (death, dementia, any level of functional disability, and level 2 or greater functional disability) and 34,181 respondents for 36 other outcomes. The primary analysis was performed using linear, logistic, and poisson regression analysis depending on the nature of the outcome, and Bonferroni correction was used to correct for multiple tests.

Compared to older adults who reported having many availability of neighborhood food stores, those reporting fewer stores were associated with less favorable outcomes in 5 of 7 domains; physical/cognitive health (low self-rated health and Instrumental activities of daily living), health behaviors (low frequency of going out), mental health (more depressive symptoms and hopeless), psychological well-being (low happiness and life satisfaction), cognitive social capital (low community attachment). These associations remained statistically significant after Bonferroni correction (p < 0.0013).

Improving the availability of neighborhood food stores may promote health and well-being.

The analysis of COVID-19 mortality revealed that the Brazilian population was critically impacted by the pandemic. However, many knowledge gaps remain regarding COVID-19 morbidity in the country. This article aims to analyze the consequences of the coronavirus disease on health situation of the Brazilian population.

This was a cross-sectional epidemiological online survey using an electronic questionnaire between July and December 2023. The sampling method used was the virtual Respondent Driven Sampling (RDS). Changes in socioeconomic conditions were assessed in the post-pandemic period. Anti-COVID-19 vaccination coverage, prevalence of SARS-CoV-2 infection were estimated, as well as of sequelae lasting three months or more (Long COVID) among confirmed cases. Associations of Long COVID with self-reported heath status, sleep disorders, and depressive symptoms were analyzed.

The sample included 3805 individuals 18-years or older. Regarding vaccination, 61.5 % (95 % CI: 58.0 %-65.0 %) stated they had received 3‒4 doses. In the post-pandemic period, 41.6 % faced financial difficulties. Prevalence of confirmed SARS-CoV-2 infection was 40.2 %, 6.4 % of respondents reported having had COVID-19, although not confirmed by test, and 15.3 % did not know if they had been infected with the coronavirus. Among those infected, 32.0 % (95 % CI: 28.8 %-35.3 %) reported Long COVID, 21.4 % reported a COVID-19-related illness, and 5.2 % needed and obtained hospitalization. Long COVID was associated with worsening self-rated health, sleep disorders, feelings of depression and 27.7 % were unable to perform their usual activities for one month or more.

The results of this study showed that Brazil was severely affected by the COVID-19 pandemic, both in terms of mortality and morbidity. The availability of timely post-pandemic data, as presented in this study, may be highly relevant to inform public policies aimed at promoting healthy behaviors, controlling NCDs, improving mental health care, and supporting specialized care for Long COVID within the public health system.

The October 7, 2023, terrorist attack in Israel resulted in widespread psychological distress, significantly impacting mental health at a national level. This study examines the longitudinal relationship between depression and suicidal ideation (SI) in the aftermath of mass trauma, with a particular focus on the moderating role of loneliness. Given prior research demonstrating the link between depression and SI, this study seeks to explore whether loneliness exacerbates this association over time, thereby informing targeted suicide prevention efforts.

This study employed a nationally representative prospective design with a cohort of 600 Israeli civilians (Mage = 41.02, SD = 13.79; 50.3% women). Data were collected at five time points: two months pre-attack (T1), one-month post-attack (T2), and three additional follow-ups across one year (T3-T5). Participants completed validated self-report measures assessing depression, SI, and loneliness, with pre-attack depression levels serving as a covariate. Hierarchical regression and moderation analyses were conducted to assess the interaction between depression and loneliness in predicting SI at T5.

Probable depression at T2 significantly predicted SI at all subsequent time points. Furthermore, loneliness at T2 independently contributed to SI at T5, beyond the effects of pre-existing depression and trauma-related factors. Crucially, moderation analyses revealed that loneliness T2 amplified the association between depression T2 and SI T5, such that individuals with high loneliness exhibited a stronger depression-SI link over time.

Findings highlight the compounding effect of depression and loneliness on suicide ideation in the aftermath of mass trauma. The results suggest that loneliness may act as a critical mechanism that sustains suicidality by reinforcing isolation and limiting access to emotional support. This study underscores the need for trauma-informed suicide prevention strategies that incorporate interventions aimed at reducing loneliness, enhancing interpersonal connectedness, and promoting social reintegration.

Electronic health records are essential for advancing research aimed at improving clinical outcomes. However, stringent data protection and privacy concerns severely limit the accessibility and use of real clinical data, particularly within Child and Adolescent Mental Health Services (CAMHS) involving vulnerable young individuals. This challenge can be effectively addressed through synthetic data generation, which safeguards individual privacy while facilitating comprehensive analyses of clinical information.

This study aims to investigate whether hierarchical synthetic data generators (SDGs) can effectively replicate the statistical properties, preserve the utility, and maintain the privacy of real CAMHS clinical data, thereby enabling data sharing and broader access to research-ready datasets.

This retrospective cohort study used electronic medical record data from 6924 distinct patients from CAMHS in Stavanger, Norway, comprising 7730 referral periods and 58,524 episodes of care. An 80%-20% split was used for training and testing. A hierarchical synthetic data generation model was trained to generate synthetic referral periods and associated episodes of care. Data quality was evaluated using SDMetrics for distribution (Kolmogorov-Smirnov Complement [KSC]/Total Variation Complement [TVC]), correlation (CorrelationSimilarity [CS]), and cardinality (CardinalityShapeSimilarity [CSS]) similarity. Privacy was evaluated using the Anonymeter library to simulate singling out, linkability, and inference reidentification attacks. Utility was assessed using the train synthetic test real (TSTR) pattern, comparing the predictive performance using precision-recall area under the curve [PRAUC] of models trained on synthetic vs real data for classifying the intensity of care.

The hierarchical SDG created highly reproducible synthetic CAMHS data. The average statistical similarity scores were high across all metrics: KSC/TVC at 0.92, CS at 0.77 (intertable CS at 0.75), and CSS at 0.92. The synthetic data also demonstrated a low risk under simulated privacy attacks on a control dataset (n=1546): the average success rate was 6/1546 (0.39%) for singling out and 77/1546 (5%) for multivariate attacks. The average linkability risk was 54/1546 (0.5%), and the highest inference risk for a sensitive variable was 2/1546 (0.12%). The classification model trained on synthetic data (TSTR) produced comparable predictive performance (PRAUC=0.40) to the model trained on real data (PRAUC=0.43) for classifying the intensity of care (low vs medium or higher). Shapley additive explanations analysis confirmed that the synthetic model's explanations aligned with real-world insights, validating its ability to capture fundamental predictive patterns.

Synthetic data can be used to build trust and promote collaboration among CAMHS researchers by offering access to extensive, representative samples with a low risk of patient identification. This approach expands the breadth of research while safeguarding patient privacy. Effective implementation of synthetic data generation depends on the model's ability to accurately identify and replicate the complex, sequential patterns present in real data.